Saturday, August 14, 2010

Optimizing images in 8 easy steps - the real basic basics

Sorry, this post has moved.

Click

HERE

to view it.
Posted by
Labels: , , ,
Your reaction: 

3 comments:

  1. Another terrific post, Gordo. I do have a quibble or two, though. I believe setting the dark and light points to RGB=10 and RGB=245 is better than 0 and 255. It gives a slightly flatter image but saves the lightest and darkest values between 0-10 and 245-255 that would otherwise disappear. YMMV.

    As to sharpening, rather than letting the Perfect be the Enemy of the Good, I'd suggest just setting the Threshold to 3 levels, the Radius to .5 pixels for web images and 1.5 for print images, and then changing the Amount to whatever the image requires as per your suggestions. This is usually 50-100% for the image I work with. This can greatly simplify an otherwise daunting process.

    I've been away from the industry for a decade now, and things may have changed (though I very much doubt it), but I was always struck with the "reverence" many printers held about the poor images customers furnished. Printers were very reluctant to make routine changes that would significantly improve the printed result. They had no trouble correcting mis-spelled words in a furnished manuscript, but improving an image was different. Photoshop has made the process so much easier now that, as you point out, it should be routine. It is probably the one single thing that most small printers can do to improve their quality.

    ReplyDelete

  2. Thank you for commenting.
    I don't disagree with you at all.
    Some thoughts about my reasoning. I was trying to get the notion of using numbers rather than simply relying on what is seen on the screen. Because our eyes auto white balance and assign white to what is assumed is white because of context, many people get misled by what they see and make incorrect image adjustments. The numbers are more reliable indicators as to what's going on in the image. For the highlights - I was referring to specular highlights, e.g. the reflection of the sun off of chrome, a highlight in the eye, etc. In the old days, if there were, say, 1% dots there, they would get lost in the print process and you'd end up with what you wanted - a 0% in the specular highlights. Today, with CtP and modern presses that is not always the case. That being said, your advice to set the values closer to RGB=10 and RGB=245 is well taken.

    For clarity's sake you are referring to sharpening the image using the "Unsharp Mask" method in PhotoShop. I didn't get into the different sharpening methods because that's a big subject and likely best covered in a separate post.

    The values you suggest:
    - Threshold to 3 levels
    - Radius to .5 pixels for web images and 1.5 for print images
    - Changing the Amount to whatever the image requires typically 50-100%

    Are not too far from what I used to use (I've switched to doing sharpening using the "High Pass" method)

    These are the values that I used to use:
    - Threshold to 10 levels (helps avoid noise becoming more visible)
    - Radius to .5 pixels for web images and 1.5 for print images
    - Changing the Amount to whatever the image requires e.g. 150%-250%

    I also agree that enhancing images or helping their customers to enhance their own images is probably the one single thing that most small printers can easily do to improve their quality. One doesn't have to become a PShop guru - just use a few simple tools in a consistent manner.

    ReplyDelete

  3. Gordon: Interesting post. I love your blog!
    Just a comment about one step. when you talk about target profiles I think it's wrong using "assign" in that part.
    Assign should be used only if the color profile was stripped or you're sure the right profile is the one you'll assign.
    But if your image has a profile and you assign a different one the result won't be correct, resulting in a dull or over-saturated image (depending which were the source and target profiles).
    If yo have sRGB and your working profile is AdobeRGB, you have to convert the sRGB colors to latter's colors, using a perceptual or colorimetric intent.
    That will preserve the correct appearance of colors.

    Also, the use of AdobeRGB (if the source material was sRGB) depends on the print profile. If you're using an output target with a gamut that fits in sRGB gamut, it has no use to move from sRGB to AdobeRGB.
    It won't be very useful if target gamut is larger but you'll preserve the color appearance of the source image, since all colors will be inside the source and target gamuts.

    ReplyDelete

Subscribe to: Post Comments (Atom)